FRAMEWORKS.
Three models for AI-native transformation in PE-backed environments. Built from what actually works — not what looks good in a deck.
These frameworks emerged from over 10 years of deploying AI and driving growth across PE-backed software companies in Germany, the UK, and the US. Each one addresses a specific failure mode: companies that move too slowly on AI adoption, companies that invest in the wrong sequence, and organizations that lack the internal ownership to sustain AI initiatives past the pilot stage.
By Yannic Desch
The Leaner Stack Paradox
SMBs don't need to catch up to enterprise AI. They're already ahead.
A structural advantage hiding in plain sight: the simplicity that makes smaller companies feel "behind" on technology is exactly what lets them move fastest on AI.
One database wins
Enterprises spend 18 months on data integration before AI can touch anything. A company with one CRM and one database can connect AI and ship automations in 3 days. The constraint is the feature.
No committee drag
Enterprise AI projects die in steering committees. SMBs make a decision and move. Speed of execution compounds faster than budget.
Less to untangle
14 tools that don't talk to each other is not a tech stack. It's debt. The companies that move fastest on AI are the ones with the least to clean up before they can start.
AI Compounding Framework
Most companies build their AI roadmap backwards. Three phases. In this order.
A sequenced model for building AI capability that pays for itself — from internal automation to commercial advantage. Each phase funds the next. Skipping phases destroys value.
Phase 1 — Plant
Look inward first. Every function, every process. Where is time and money going that AI could handle? Automate it. Free up the margin. You can't build AI capability on borrowed budget — self-funding changes the entire narrative.
Phase 2 — Build
Reinvest that margin into capability, not more tools. AI engineering. Agentic development. People who own AI inside your business. Tools can be copied. Internal capability can't. That's the moat.
Phase 3 — Harvest
Now build for revenue. Faster product cycles. Better margins. AI that shows up in the P&L — not just the pitch deck. Customer-facing AI features are credible at this stage because they're built on operational discipline, not bolted on.
Related writing
The Agent Orchestrator
The hire most companies are missing. Not in IT. Reporting to the CEO.
A new internal job profile that accelerates AI adoption 5x by treating AI agents like a team member — assigning tasks, managing output quality, and bridging business and technology.
Operator, not engineer
The Agent Orchestrator understands business operations deeply — not just technology. They're a builder who learned to operate, not a technologist who learned about business. The difference matters at execution.
Reports to CEO or COO
Not the CTO. This is a commercial role, not a technical one. Success is measured in EBITDA impact, not deployment count. Embedding it in IT creates the wrong incentives.
One metric: time and cost removed per week
The role exists to find, build, and deploy AI agents across Sales, Ops, Finance, and Product simultaneously. Without this dedicated owner, AI adoption stalls inside committee meetings while everyone waits for someone else's roadmap.
Applying any of these in your organization? I'd like to hear what you're seeing.
Get in touch→